LINGUIST List 4.271

Tue 13 Apr 1993

Sum: Adjectives (Part 2)

Editor for this issue: <>




Date: Sat, 10 Apr 93 12:38:18 SSSUM & NEW QUERY: ADJECTIVES
From: David Gil <ELLGILDNUSVM.bitnet>

Following is a second summary of responses to my query on
adjectives. The first part consists of 9 slightly-edited responses
to my query and/or the first summary, arranged according to
language; the second part contains some general comments of
my own, and a new query.


ENGLISH (Larry Gorbet)

Just read your summary of responses to your query about
adjectives in NP positions. Some of the English data is
questionable in relevance, I think. In particular, many (classes
of) "adjectives" that occupy N or NP position act very much like
lexical nouns, not adjectives. In particular, signs of this
lexicalization include

1. idiosyncratic number
2. idiosyncratic nominal content (i.e. the "adjective" is
interpreted only as a noun of a particular class modified by the
3. idiosyncratic senses (i.e. the "adjective" is only interpreted
with a proper subset of its possible senses)


Note that while (a) is OK, (b) is not:

(a) The carnivorous are ferocious
(b) *The carnivorous is ferocious.


(c) Lonely are the brave.
(d) *Lonely is the brave.

(e) *The obvious are not always correct.
(f) The obvious is not always correct.


Lots of "adjectives" only refer to humans despite in principle
being applicable to other classes of nominals

(g) Only the tall will make the team. [humans]
(h) ?*Only the tall will be preserved. [trees]
(i) *Only the tall are unstable. [buildings]

Others, like (f) above, only apply to abstract propositions


(a) I'll take the large. [nominal clothing size]
(b) *I'll take the large. [actual size of "unsized" shawl]

The distribution of these supports the fact that they are
lexicalized nouns. That is, the distribution of particular
"adjectives" and of certain semantic (etc.?) classes of
"adjectives" fits the nominals (and senses) with which one
would expect them to be used most often.


DUTCH (Jan Rijkhoff)

In response to your query some time ago on the use of dummies
in terms "headed" by adjectives: I believe that might be a
correlation with (grammatical/noun inherent) gender. For
instance, Dutch (gender): de rode (the red one), de grote (the
big one), de lelijke (the ugly one); notice that English (no
gender) has the dummy.
A short time after I thought of this possible correlation
(several years ago), I read an article by Pieter Muysken and
Frans Hinskens, who seemed to have had the same idea (more
or less), so I let it rest. If you are interested I can send you a
(non-final) copy of their article (in Dutch I'm afraid), which
I must have somewhere in my "archives".


GREEK (Stavros Macrakis)

There appear to be three issues here, and it's not clear which one
you're interested in:

1) Can you use an adjective without a noun to form an NP?

2) In this case, do you need to mark the adjective in some special
 way, different from a noun?

3) Is there some sort of "pronoun" like "one" which takes the
place of the noun.

In Modern Greek, you have (1) but not (2) or (3):

 Pjo fustani protimas?
 Which(n) skirt(n) prefer(2s pres. ind.)?
 Which skirt do you prefer?

 To kokino.
 The(n) red(n).
 The red one.

 To kitrino ine fthinotero.
 The(n) yellow(n) is cheaper(n).
 The yellow one is cheaper.


HINDI (Bhuvaneswari Narasi)

 I just saw the responses to your request for examples
of adjectives that occur as complete NPs'. Here's an
example from Hindi:

 "Lal gadi" = red car

 "Lal vala" = the red one

"Vala" (or "wallah" as the British spell it), also
attaches to various nouns as an agentive suffix -
e.g. "dhobi-vala" (washerman). With adjectives however,
it has the meaning "one".....


ARABIC (Maher Awad)

[...] regarding your questions of adjectives occupying NP slots
by themselves,
Arabic has constructions equivalent to:
I want the red. (meaning I want the red one)
I want a red one. same meaning.
But you cannot have:
I want the red one.
In short, you cannot have 'one' with an indefinite article (zero in


TURKIC (Vern M. Lindblad)

The common wisdom about Turkic langs. is that there are two
sets of stems, nominals and verbals (leading to the Turkological
convention of using + to mark nominal morpheme boundaries
and - to mark verbal morpheme boundaries). There are very few
stems that can take both sets of suffixes (though there are
various suffixes that change nominals into verbals and vice
versa). However, many 'nominal' stems quite freely take both
nominal and adjectival suffixes, so that there is only minimal
differentiation between nouns and adjectives. Thus, it probably
easier for an adj. in a Turkic lang. to stand alone as an NP than
in most other langs. However, this is not something that I've
looked into seriously, so my comments should be taken as
suggesting a possible line of inquiry, rather than as authoritative.


JAPANESE (Bart Mathias)

As a specialist in Japanese, I refrained from responding to your
question about "red one" type languages because, as you
surmise on the basis of the notes you got on Japanese, the "no"
in question is a nominalizer, but not a noun (or even a word,
though I know of one problem that gets in the way of calling it a
suffix) apparently of the type you mention in your original

I'm just writing this to confirm that surmisal, because your
correspondents did not make it quite clear. "No," unlike "one,"
is not an independent word; it must follow--attach to--a tensed
form or a noun(! the situation transformationalist types like to
consider a reduction of "no no," where the former "no" is a
genitive, but that doesn't help much). When it follows a
tensed form it can be thought of as the head of a relative clause
(thus closer to "one that is red" than to "red one"), but in fact it
also ends "headless relatives," and should be considered as
giving a syntactic role to the word it ends rather than playing its
own syntactic role.


ASL (Therese Shellabarger)

I am a second language user of ASL, so don't have to take my
word for this exactly, but in ASL one can sign using what
English uses as adjectives for nouns, and specifying quantifiers
so that you could sign an equivalent of "I want the Red one"
with "red" being the noun and "one" being how many, only if
there were more than one "Red" objects of the same kind to
pick from. Red would be used as a distinguishing feature of the
object in question. If you used "one" simply to match the
equivalent English sentence, then you would be using signed
English, which gets into a totally other ballgame. As an example
of the above, a deaf woman I used to live with pointed out to me
the good ASL of her children one time to me when they were
cold and wanted the "orange hot"--meaning, the heater which
when hot the elements glowed orange. My gloss of their signs is
not meant to be taken as an English translation, but rather to
clarify what they signed for readers who know ASL...


GENERAL (Kate Kearns)

I just read your summary of responses to your query. I
felt people were responding with two different types of
phenomena because there were two types in your query:
NPs with dummy heads, as in your example "I want the red
one", where the dummy is anaphoric or otherwise contextually
identified, and "adjectives standing by themselves as NPs",
maybe with some morphology, but presumably distinguished
from the other construction by the absence of a separate nominal
element as head of NP. A candidate for the second type in
English would be 'The good, the bad and the ugly', 'Eat the
rich', (movies) or 'the poor are always with us'; these don't
have any context-dependence. Maybe there is an empty category
bound by 'the', but it's not clear how it is governed if the whole
NP is in subject position. Also 'The secret of life is taking the
rough with the smooth, the good with the bad' - here I think
'good' and 'bad', presumably also 'rough' and 'smooth', strike
me as zero conversions to nouns.



My original impetus in posing the query was to check out a
hypothesis that was suggested to me by Martin Haspelmath, and
is similar to that proposed by Jan Rijkhoff above, namely that
the ability of an adjective to stand alone as head is correlated
with the richness of adjectival morphology in the language in
question. That is to say, if an adjective bears lots of nominal-
like inflections, it will be able to head an NP, whereas if it is
morphologically bare, it will need some kind of dummy "one",
nominalizer, or other such prop.

The results of the query, and my own work, cast some doubt on
this hypothesis, while perhaps supporting a modified version of
it, involving a uni-directional implicational universal. The
following table provides a very rough classification of languages
in accordance with the richness of their adjectival morphology
[columns] and their strategies for letting an adjective constitute
the main semantic element of an NP, ie. for saying "(I want the)
red one" [rows].

 little or no A rich A
 morphology morphology

bare A Hungarian, Estonian Dutch, Hebrew
 Malay (Peranakan) <Punjabi>
 Tagalog, <Punjabi>

A plus English, Sinhalese
dummy "one"

A plus Mandarin, Japanese <Punjabi>
nominalizing Malay (Standard) (particle)
particle or <Punjabi> (particle)
affix Lezgian, Malayalam

Note that the top left corner cell, containing languages with little
or no adjectival morphology but nevertheless allowing adjectives
to head NPs, is well-documented: these languages thus violate
the proposed universal. However, the middle and bottom right
cells are still rather weakly attested; so there may perhaps be
some basis for a uni-directional implicational universal, to the
effect that if a language has rich adjectival morphology, then it
will permit its adjectives to head NPs (but not vice versa).

(The only counterexample that I am familiar with to the latter,
weaker claim, is Punjabi, which, as suggested by the angular
brackets, simultaneously fills four out of the six cells. What this
means is as follows: in Punjabi there are two classes of
adjectives, one with gender marking, the other without; and
*both* classes can either occur as bare heads of an NP or in
construction with a nominalizing particle "vaala", similar to
Japanese "no", Mandarin "de", Malay "yang", etc. Thus,
Punjabi singlehandedly refutes any correlation between
morphological richness and the ability of an adjective to stand
alone as NP head.)

So here's a more specific query: can anybody provide better
examples of languages to fill the middle and bottom right cells?
That is to say, is anybody familiar with a language that has rich
adjectival morphology, but in which adjectives cannot stand by
themselves as heads, but need a dummy "one", or nominalizing
particle or suffix?

Finally, a methodological/philosophical observation. At the
same time that I posed the original query on adjectives, a rather
heated discussion of pro-drop was taking place over the list.
Although there are some interesting parallels between the two
issues (they both concern the licensing of "empty" positions; in
both cases correlations with morphological richness have been
proposed), the adjective query didn't generate anywhere near as
much interest as the pro-drop issue. These days, it would seem,
verbs are just plain sexier than adjectives. Any ideas why?
(And is this a fact about verbs and adjectives, or a fact about us
Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue